Need replacement PSU, Questions inside.

[Antimatter]

Alright here's the deal, my Antec 450 Watt PSU, I'm suspecting that it ate two of my hard drive and soon an 3rd is about to fail.

And I'm not sure how to track down what is eating up my hard drives? Its bit suspect because I had one hard drive go bad, so I yanked that one out, a few days another is starting to go bad, so I left it in for a while, while i was dealing with RMA stuff, then finally yanked it out, now a few days after i yanked out that 2nd hard drive my 3rd hard drive performance is starting to go down hill from 60MB/sec to about averaging 20 MB/sec and already lost it once in the BIOS and had to do an hard restart (unplugging the power supply and replug it in) to get it back.

Plus in the last year once in a while i would mysteriously lose one hard drive in windows and the BIOS, then having to need to do a hard restart to get them to reappear again.

Anyway the spec of the system is following.

Antec 450 Watt with following spec.
3.3 volt - 32 amp
5 volt - 40 amp
12 volt - 24 amp

System spec:
IC7-max3
4x 512 Mb corsair memory stick
sapphire 9800 pro 128 mb
Intel gigabyte network card
haupapage 250 PVR tv card

fans:
2x 120 mm tornado
4x 80 mm tornado

disk drive:
1x dvd burner - pioneer
1x cdrom burner - liteon

2x 200 gig western digital hard drives - already RMA
2x 36 gig 10k Raptors - one of the hard drive performance is starting to drop in performance, and suspect its going to fail soon.


Anyway to check the estimated power requirement i went ahead and used the Takaman's power supply calculator, which came out with this following spec. And I'm a bit suspect of the 3.3 volt rail requirement though.

3.3 volt - 2.7 amp or above
5 volt - 32.4 amp
12 volt - 24.4 amp

And compare it to my PSU amp rating.

3.3 volt - 32 amp
5 volt - 40 amp
12 volt - 24 amp.

So it appears that the source is probably my power supply unit considering the power supply calculator estimated the requirement at 24.4 amp for the 12 volt rail, and the PSU coughs up 24 amp for 12 volt rail.

So I went on out to search for an PSU that had something similar to this for its minimum amp rating

3.3 volt - 25+ amp
5 volt - 35+ amp
12 volt - 30+ amp

And I haven't had much luck, the most powerful fortron and sparkle that i found had 29 amp for their 5 volt rail, so those probably are out. which leaves the zippy and the entermax, and i prefer to have two fan for my PSU because of the layout of my computer (its custom modded) so the zippy may be out, which leaves the entermax.

Zippy
Enermax
Fortron
Sparkle


So got any ideas/advices here?

Plus on top of that all, i don't have any room above or below my 20 socket motherboard plugin for an 24 socket, so that is pretty much out of the question, and it also needs the 4 socket aux power.

 

[larrymoencurly]

Takaman's power estimator may be the most accurate around, but even it gives overly high estimates, and another person with a similar HD setup and an Athlon64 3000+ or 3200+ found that his system ever drew more than 400W -- from the wall outlet, meaning that the PSU probably put out only 300W max. IOW I think that Takaman's tells you how powerful a decent quality PSU you need (and Antecs are much more than decent), not what the computer will actually consume. It's only with the +3.3V amps estimate where Takaman's gives lowball figures, often by a factor of 2-3x.

If the voltages are OK (well within 5%) when measured with a digital multimeter (mobo readings aren't nearly as trustworthy as those from a $5 digital meter), then the the PSU is almost surely powerful enough. Even better, seek out a techie friend who owns a DC clamp-on ammeter or probe because with it you can measure the amps on each rail. I wouldn't expect more than about 10-15A flowing from the +5V, and the +12V will probably be putting out only 15-20A, and that's assuming a separate +12V power connector for the 9800 Pro graphics card. However a fast dual-CPU mobo can be a real power hog.

What CPU(s) do you have? Also how are you installing the HDs? Do they have at least 0.5" of air space between them?

 

[Antimatter]

I have the regular Pentium 4 at 2.8 ghz and 800 FSB.

And I'm using an lian-li case, and in front of my hard drive bay is two tornado at about 80 CFM fans so there's a bit of air moving though, and the way its set up is.

| - Wall
@ - Air gap
R - Raptors
A - Regular IDE

| R @ R @ A A |

So in other words theres approx 1-2ish inch gap between each of the raptors. Then due to the fact that the IDE are only 7200 RPM thus probably cooler and the fact that the lian-li hard drive bay only holds 5 hard drive i had to bunch them together at the end of the tray but they have at least half a inch gap between each one.

About the multimeter, I think i have one around, so can i just stick it into any old socket and see what sort of voltage they are getting? But before I do that I'll check the wiring diagrams ofc.

And i think that my meter also supports amps, checked, its only good to 250 mA so that's out of the question.

 

[Ice Czar]

as larrymoencurly points out alot of amp figures get padded, especially in CPUs where more often than not its a max thermal rating in watts they are derived from, and videocards are notorious for being secretive on their power draw

the one area where a little digging can produce good figures is HDDs
there you can often find both the max draw for spinup, an "idle" draw (inactive but spinning) and an incrementally higher draw for a seek where the armature is being powered

more common in enterprise class drives than those aimed at the end user

there are very expensive multimeters that will measure amp draws in DC with RMS through an inductive clamp, but odds are yours isnt one of them, thus youd be observing just voltage fluctuation on the rail, not the actual amp draw of a given drive

 

[Antimatter]

as larrymoencurly points out alot of amp figures get padded, especially in CPUs where more often than not its a max thermal rating in watts they are derived from, and videocards are notorious for being secretive on their power draw

the one area where a little digging can produce good figures is HDDs
there you can often find both the max draw for spinup, an "idle" draw (inactive but spinning) and an incrementally higher draw for a seek where the armature is being powered

more common in enterprise class drives than those aimed at the end user

there are very expensive multimeters that will measure amp draws in DC with RMS through an inductive clamp, but odds are yours isnt one of them, thus youd be observing just voltage fluctuation on the rail, not the actual amp draw of a given drive

Yep that's correct but i think my dad may have one so i can ask him, because I've recalled seeing an meter with a large clamp on top of it.


Any other way of determiting what the possiable cause of failure is, because i think its highly suspect that 2 harddrive failed within a short time peroid, and 3rd is starting to degerate also.

 

[Ice Czar]

Yep that's correct but i think my dad may have one so i can ask him, because I've recalled seeing an meter with a large clamp on top of it.

ahh, we are now into the tools that I covet

a Fluke 189 multimeter
generally when you see an inductive clamp on a multimeter, its rated for AC measurement
more rare are those that will do DC w\ true RMS at a reasonable accuracy.

there are quite a few that will do DC current without RMS though I gather

 

[Antimatter]

ahh, we are now into the tools that I covet

a Fluke 189 multimeter
generally when you see an inductive clamp on a multimeter, its rated for AC measurement
more rare are those that will do DC w\ true RMS at a reasonable accuracy.

there are quite a few that will do DC current without RMS though I gather

Ohh that thing, I've seen a couple floating around and always would have liked to own one of them But i don't think my dad model is one of those but I'll ask him, he's being little pissy tonight over something, so I'll ask tomorrow.

But an update on my computer, the hard drives seems to be stable right now, but the performance of one of my sata hard drive has dropped to 20 Mb/sec now for read performance, but it has so far stopped degerating, and haven't had any other issue so far with power rails being unstable or anything, so now with two hard drives out of the picture, the power draw is probably lower thus the rails are stabler, but I'm still suspectious of the power supply.

 

[Antimatter]

Oops I just now remembered that i also had other source of power draw.

4x cold cathoide tube, each about 1 1/2 foot long... will need to look the spec up on em

1x johnson marine pump, i think it takes 12 volt, and its an pretty hefty pump, so that's another source of power draw that i also forgot to include in the power draw table.


I'm going to start to dig around for some detailed spec and hopfully be able to draft up some detailed current draw based on the system spec i have, but i have a suspection that my pump and those cold cathoide lights are also what is pushing the power draw higher than normal.

 

[Ice Czar]

cathodes are notorious for $#$@#UP systems

The inverters for cathodes are horrible. I'm never installing another cathode unless I can come up with a way to where they have their own power source. The inverters dirty up the rail and cause horrible voltage fluctuations.

once they are warmed up they are around 4W per, before that...

 

[Ice Czar]

cathodes are notorious for $#$@#UP systems

The inverters for cathodes are horrible. I'm never installing another cathode unless I can come up with a way to where they have their own power source. The inverters dirty up the rail and cause horrible voltage fluctuations.

once they are warmed up they are around 4W per, before that...

Ive never used cathodes, not much point there are no windows in the rackmounts

 

[Antimatter]

cathodes are notorious for $#$@#UP systems



once they are warmed up they are around 4W per, before that...

Ive never used cathodes, not much point there are no windows in the rackmounts

Interesting, well i have a seperate 12 volt rail for my pump, aka my own power source for the pump so maybe i can move my cathoides on over to that power source instead of the computer?

 

[davidhammock200]

Interesting, well i have a seperate 12 volt rail for my pump, aka my own power source for the pump so maybe i can move my cathoides on over to that power source instead of the computer?

Good idea.

 

[Antimatter]

Good idea.

Atleast it'll keep as much "non-computer" stuff off the computer PSU and on the alternative power supply unit.

I would love to have one of those pcp&c power supply unit, but daaamn theyre frigging expensive! But i'm thinking of going ahead and bitting the bullet and picking up one of those, mainly for furture expansion flexability, and also more reliability/guarate that i won't lose another batch of shit to the psu.


Man what a pain, I went out and searched all of the technical documination and specification that i could find for all of the devices on my system, and i picked their max draw such the harddrives spinup of 1.9 amp for some of em, but some didn't have any spec so i had to make estimation/aprox on those, but for most part after i was done my estimation and math was aprox 17% over or under the takaman's power supply calucator.

 

[renegade44e]

I have an enermax EG565P-VE powering

ABIT NF-7S
2500XP-M @3200 (2200MHZ)
2X 5i2MB Kingston HyperX 3500
ATI 9800 hacked to pro
Audigy 2 Platinum
PVR250MCE TV Card
PVR150MCE TV Card
Promise SATA S150 raid card
BenQ 1620 DVD burner
MSI 16X DVD-ROM
3 WD2500JD
3 Maxroe Diamondmax 9 sata HD

Yes that is 6 X 250GB 7200RPM Hard Drives!!!

And 4 fans in a silverstone TJ-06

 

[Ice Czar]

pumpds are a pretty static draw once you get them spinning, but non-computer components, and especially the often cheap inverters on cold cathodes often lack good sheilding and add all sorts of nasty harmonics to the power from what I gather and have been linked empirically to alot of issues

Im hoping jonnyGURU gets an oscilliscope soon so we can determine exactly what a cold cathode is capable of doing to a line

the basic spec (ATX12V) calls for 120mV AC Ripple on the +12V rail(s)
some supplies do much better than that